According to the latest report, GPUs will jump directly from 28nm to 16nm and TSMC's 16nm FinFET+ manufacturing process should offer up to 40 percent increase in performance at the same power draw and should go into volume production in the third quarter of this year.

While we already knew that GPUs will skip 20nm manufacturing process and go directly from 28nm to 16nm, a recent report from Cadence publication, spotted by Wccftech.com, reveal a bit more details regarding TSMC's 16nm FinFET+ manufacturing process as well as new 28nm HPC+ and 16nm FFC manufacturing processes aimed at Internet of Things (IoT) market.

According to the new report, TSMC's 16nm FinFET+ manufacturing process will offer around 40 percent performance increase at the same power draw or, basically, draw 50 percent less power at the same performance level. Currently, 16nm FinFET+ manufacturing process is approaching 20nm yields and looks quite good, even better than any preceding TSMC processes, according to the report. Most GPUs will be based on 16nm FinFET+ manufacturing process although, both Nvidia and AMD have only announced that next-generation GPUs will be based on FinFET manufacturing process and did not specify if it will use 16nm TSMC or 14nm Samsung manufacturing capabilities.

In addition to the TSMC 16nm FinFET+ details, the same report also gives details on two new low power and mobile manufacturing processes aimed at IoT market, the 16nm FinFET+ Compact (16nm FFC) and 28nm High Performance Compact Plus (28nm HPC+) manufacturing processes.

The 28nm HPC+ will offer an upgrade to the existing 28nm HPC process by providing up to 15 percent more performance at the same process leakage, while 16nm FFC manufacturing process is expected to enter risk production in 2016 and have first volume ready by 2H 2016.

Since GPUs have been stuck on 28nm manufacturing process for a while it will be quite interesting to see what the 14/16nm FinFET manufacturing process will bring, or will manufacturers have the same issues as with 28nm manufacturing process and what will be the actual performance gain when it comes to flagship GPUs.



Source: Wccftech.com.